Description and overview of dataset:
Comprehensive mapping of burn sites in the West Line, Palestine (1999-2023).

Generated on 2024-03-16

Recommended citation for the dataset: 
Garb Y., Leblond N., Tatweer wa Nahda, Alswaitti A., Awawdi S., Alyamany F., Alhroub A., Sharawnah I., Um Muhamad, Alswaitti M., Davis, J.M., Alswaitti Sa., Alwaitti Su., 2024, Comprehensive mapping of burn sites in the West Line area (1999-2023). Version 2024-03-10.  DOI:  https://doi.org/10.7910/DVN/VUKLGP

**********************
CONTEXT AND OBJECTIVES 
**********************

1. This burn site mapping dataset is the product of a decade of our team�s effort to understand and document the phenomenon of waste burning in the West Line area of south-west Palestine, document the environmental and health impacts, and support alternative policies and practices such as livelihood alternatives, community-based burn enforcement, and contaminated site remediation.  A bibliography listing the relevant project scientific publication as media coverage is appended to this report, and the materials themselves are available from the project PIs upon request. 
2. This dataset represents the manual identification and tracing of burn sites in the southern West Bank using high resolution aerial images (resolution ranging from 10 cm to 50 cm), validated through careful ground-truthing of a subset of the sites, and systematic quality control(iteration of inclusion criteria, and several steps of checking and independent cross-correcting by multiple team members.  The mapping process took place within a broader research envelope of hundreds of field visits, the sampling of over 40 burn cores for contaminant analysis, and numerous interviews and meetings with local authorities, e-waste related businesses, waste burners, concerned inhabitants, and local associations. 
3. This repository is intended to be a dynamic resource, evolving over time to serve research and policy for the affected areas and other similar areas globally. The initial release described in this document is of twenty two images showing aggregated data covering the two decades (1999-2023) of the burning phenomenon on an almost annual basis.  These images are heat map (kernel density) representations of the density of thousands of burns.  In addition, we have superimposed on this derived density layer a point layer of the burn sites location after the addition of enough random noise to preserve fidelity to overall patterns while providing enough spatial anonymization to protect the particular residents and landowners of the affected areas.
4. Because it is based on many thousands of person hours of informed site identification by people very familiar with the landscape, as well as extensive ground validation, including of edge cases, this data repository of burn sites represents a rich and high-fidelity platform for further analysis. Our team is now building upon this repository as well as other data layers and surveys, as the basis for analyses and policy recommendations related to the spheres of health, agriculture, remediation, contaminant transport and hydrology, etc.  We welcome collaborations with specialists in these areas, as well as experts in remote sensing and spatial analysis techniques that would like to join us in using the raw data as a training set for improved and automated detection approaches, including multi-sensor, multi-spectral, and.  Get in touch with us to see how we could work together. 


*******************
PROJECT INFORMATION
*******************
1. Title of dataset: 
Comprehensive mapping of burn sites in the West Line area (1999-2023). 

2. Author information

	Principal Investigator
	Name:Prof. Yaakov Garb
	Email: ygarb@bgu.ac.il 

	Co-Principal Investigator
	Name: Dr. Nelly Leblond
	Email: Nellyleblond11@gmail.com 

	Director of Tatweer wa Nahda
	Name: Sara Awawdi
	Email: Tatweer_nahda@yahoo.com 

	Co-researchers (2012-2023)
Name: Alswaitti A. (Tatweer technical lead), Awawdi S., Alyamany,  F., Alhroub A., Sharawnah 	I., Um Muhamad, 
	Address: Tatweer wa Nahda, Deir Samit, Hebron, Palestine,
	https://tatweer-nahda.org/ 
	Email: Tatweer_nahda@yahoo.com 

	Co-researcher (1999-2011): 
	Name�: Ass. Prof. John-Michael Davis
	Email: jdavis4@wpi.edu 
	
3. Date of data collection 

All the heatmaps are derived from the analysis of high-resolution aerial photographs collected, corrected, and consolidated by Y.  Garb (2010-2024). Analysis was performed by the authors (see Methods section) between 2011 and 2013 for the 1999-2011 images and between 2023 and 2024 for the 2012-2023 images. 

Y. Garb and JM. Davis co-devised a protocol to map e-waste burn sites for the 1999 to 2011 period and Davis manually identified the burn sites under his supervision. Field validation throughout the Hebron governorate for this period was performed by JM. Davis with the support of S. Awawdi, S. Alswaitti and Y. Garb.

In 2022-2023, Y. Garb and N. Leblond adapted the protocol and trained a team of mappers with the support of A. Alswaitti at the association Tatweer wa Nahda for the analysis of the 2012 to 2023 images. The high resolution images were mounted to a PostGIS server in August 2023 (setup by Luna GeoSpatial), enabling simultaneous mapping and editing. 
A. Alswaitti and M. Alswaitti set up the computers and A. Alswaitti coordinated the mapping team (2022-2024). 
N. Leblond consolidated and cleaned the files and set up the upgrading protocol with Y. Garb (2023. A. Alswaitti and N. Leblond (2023-2024) upgraded the database, correcting the omissions, false commissions, and incorrect boundaries of over 20% of the features. The heatmaps were derived from the dataset and produced with QGIS. Extensive field assessment of burn sites was conducted by Y. Garb with JM Davis (2011-) and Leblond (20XX-2024) with extensive support by and input from the entire team and members of the West Line community.

4. Funders (analysis of 2012-2023 coverage)
   	Name: Middle East Research Cooperation Program/USAID.  Grant No. SIS70022GR0016
   	Name: Fischer Fund (private donation)
   Additional funders (analysis of 1999-2011 coverage): UNFAO, Sida, Swiss Embassy in Israel.

************************
DATA ACCESS INFORMATION
************************

1. Licences / restrictions placed on access to the dataset
CC-BY-NC-SA

If you want to use this data as part of your work please get in touch with us. We can provide further methodological insights into the formation and limitations of the materials, as well as detailed underlying data, and are interested to partner to cross-analyse our data.


2. Links to publications that directly use the 1999-2011 portion of the data 
   
	Reference: Davis, John-Michael (2017) Defining an equitable Israeli-Palestinian e-waste economy. Doctoral (PhD) thesis, Memorial University of Newfoundland.
	DOI: http://research.library.mun.ca/id/eprint/13003 

Reference: Davis JM, Garb Y. A strong spatial association between e-waste burn sites and childhood lymphoma in the West Bank, Palestine. Int J Cancer. 2019 Feb 1;144(3):470-475. doi: 10.1002/ijc.31902. Epub 2018 Oct 22. PMID: 30259977.
DOI: 10.1002/ijc.31902

	Reference: Davis J-M, Garb Y. Polluted Politics: The Development of an Israeli-Palestinian E-Waste Economy. Cambridge University Press; 2024
	DOI: https://www.cambridge.org/core/books/polluted-politics/C5DBC45480C1DF14252763719BC28799

Additional references on the burning phenomena are listed at the end of this document.


***************************
METHODS OF DATA COLLECTION
***************************

1. Describe the methods for data collection

Burn sites were mapped manually based on visual identification, including the burn core itself (area on which burning took place directly)as well as immediately adjacent areas visibly affected by black ash (proximal runoff, tire-borne dispersal along roads, pushing or dumping of burn core material to the side of the core. 
The dataset combines two 2 mapping phases corresponding to 2 different periods (1999-2011 principally mapped by JM Davis under the training and supervision of Y Garb) and (2012-2023 principally mapped by the team at Tatweer wa Nahda under the training and supervision of Y. Garb. and N Leblond).  The extents, goals and procedures of these efforts differed somewhat. 
- The mapping of the first period was of the entire Hebron Governorate (2016 km�), and the second period was of the area of most intense burning within this, namely a 20 by 20 km square to the west of the Governorate, the 20% of the area containing 71% of the burns.  The maps in this data release are of this 400 km� square that was mapped for the entire 1999-2023 period. 
- For the 1999-2011 period, the mapping was performed with the objective of assessing remediation needs, hence a polygon was created covering the maximal extent over the entire period of the burn cores detected at each site.  For each year a surety score and portion of maximal of this maximal polygon was given.  Centroids were derived from the resulting layer, based on the maximal burn site polygon. The focus in this period was on burns in agricultural and residential areas. Potential burn sites in industrial settings, especially in dismantling yards, where cables are sometimes burnt for copper extraction were not included.  A separate mapping and analysis of all e-waste related businesses was done, that includes these locations. (See Davis, John-Michael, and Yaakov Garb. 2018. �Quantifying Flows and Economies of Informal E-Waste Hubs: Learning from the Israeli�Palestinian e-Waste Sector.� The Geographical Journal.    https://doi.org/10.1111/geoj.12275).
- For the 2012-2023 period, the mapping was performed jointly with the Tatweer team, with supervision of Garb and Leblond,  with the objective of assessing the health and environmental impacts of waste (both e-waste and household waste) burning. Here, for this key 20 km by 20 km area, a separate polygon was traced for the burn core extent in each year, with separate centroids derived for each year.  In addition, the greater spatial resolution of remote sensing imagery available for this later period, and the broadened goals yielded a dataset that is as exhaustively complete as possible. Thus, this period includes burns within industrial settings (typically scrap yards, and we were able to map small backyard fires where household residues are burnt.
- While the two datasets of these two periods do form a useful continuous whole with respect to the larger burns outside of industrial settings, these methodological differences must be considered before making inferences that span the two periods.  In addition, as part of our effort toward creating a unified dataset through spatial-spectral normalization of the multi-sensor and multitemporal coverage, we will be re-analysing the first period using the same criteria as the second.

2. Provide links to papers describing data collection methods
	Paper DOI: 10.1002/ijc.31902 (for the 1999-2011 layers)	
	Paper DOI: 2012-2023 layers : publication in process

2. Data processing methods and quality control

As described in Davis and Garb (10.1002/ijc.31902), before embarking on this mapping effort, we validated the remote sensing approach to burn site mapping through a cross-checking of a subset of sites to make sure that sites seen in the field were captured in the aerial imagery scanning, and vice versa.  This paper describes our methodology for the first period (1999-2011) in detail, and the methods for the second period are described below.  The dataset for the first year was reorganized by year to obtain a format similar to the 2012-2023 dataset.

As with the first period, in the second period, too, a surety score was given by the mappers to each identified site, as to boost the reliability of the dataset given the subjective nature of burn site identification.  The surety score ranges from 1 (maybe a burn site) to 3 (certain it is a burn site), with added observations in a �remark� field. In a second pass, another checker went through all the sites and (1) assigned a surety score and remark, (2) corrected the polygons boundaries as needed, (3) checked the area overall for omission, identifying sites that might have been overlooked/forgotten by the mapper. 

All the years were mapped separately (except 2023 which was mapped with previous years as guides). This �blinded� mapping of all but the last year was more time consuming but ensured a renewed and careful analysis of the area. Duplicates were removed, geometries corrected, and several targeted interventions were performed to improve the quality of the dataset: 
- Particularly large sites (>200m�)were all systematically checked. Erroneous detections due to poor image quality (confusion caused by shadows in valleys)were removed. 
- The 2012 layer (first year mapped by the mapping team and methodological change) was particularly well checked to facilitate the integration with the 1999-2011 database. We focused on places identified as burn sites in 2011 and 2013 but not 2012 to address discontinuities. 
- The 2019 layer was of inferior visual quality, with a limited contrast making it difficult to identify burn sites.  This was systematically checked for omissions on sites identified in 2018 and 2020 but not in 2019. This prior information from the prior and following year boosted our sensitivity for this problematic coverage, and enabled us to fill numerous gaps. 
- The number of burn sites detected in 2018 was much higher than in the other years, largely due to a very good image quality (high resolution, clear image). In 2018 1,886 burn sites were identified, versus 1,018 in 2017 and 881 in 2019. To make the dataset temporally coherent, for the purposes of the maps in this release, which emphasise year-on-year comparison, we removed the smaller sites (<25m�) that would have been less detectable in other years.  These mostly correspond to small household waste burning sites in front yards or backyards.
   
In total, the 1999-2011 dataset contains 1455 centroids of yearly burn sites in the 20 km by 20 km area while the 2012-2023 dataset contains 12,692 centroids in this same area. 


3. Analysis and visualization methods 

Heatmaps were generated using the Styled Heatmap (Kernel Density Estimation) in QGIS, with a cell dimension of 10 m, a kernel radius of 500 m, a Discrete interpolation and a Continuous mode. 

A random noise was applied to the burn sites overlaid on top of the heatmap. Thus, the points in the layer overlaid on the heat map layer are at some uniformly-distributed upper-bounded random distance from their true location.


4. Limits of the current dataset

Despite all the care put into mapping and improving our dataset, several limits must be considered when interpreting or using this dataset: 

- The images were generated by different sensors, at different moments of the year and times of day, with variations in terms of resolution, inclination, vegetation growth stage, and overall luminosity. The images in this release are not based on standardised or cross calibrated data at this stage, which can influence the interpretation of burn site detection and extent(see corrections above). 
- The 1999-2011 layers do not contain burn sites in dismantling facilities, while the 2012-2023 layers do, hence the 2011/2012 discontinuity is partly due to this methodological change (rather than a surge in burning). 
- Despite the removal of the smallest sites in the 2018 layer caused by the overdetection of small household backyard/frontyard burning, 2018 still stands out as a year particularly rich in burn events. The temporal pattern must be interpreted with care. 
- The material burnt on site can hardly be identified from high resolution images � hence the burn sites mapped are a combination of e-waste, household, and industrial waste burning. Contextual and grounded analysis is necessary to categorise and further assess the sites.


*********************
SUMMARY OF DATA FILES
*********************

1. List of data files
The initial deposit is of 22 files (Filenames: Heatmap1999 to Heatmap2023) containing approximately annual maps of burn sites and burn site density derived as described above from the burn site centroids.



***********************************************************
Additional references on contexts of the burning phenomenon
***********************************************************


New York Times. (September 12, 2019). �The toxic trash that is poisoning the West Bank.�

Nature. (November 30, 2016). �Can old computers bring Palestinians and Israelis together?�
Davis, J. M. & Garb, Y. (2020). Toward active community environmental policing: Potentials and limits of a catalytic model. Environmental Management.

Davis, J. M. & Garb, Y. (2019). Extended responsibility or continued
dis/articulation? Critical perspectives on e-waste policies from the Israeli-Palestinian case. Environment and Planning E: Nature and Space, 2(2), 368-389.

Davis, J. M. & Garb, Y. (2018). Quantifying flows and economies of informal ewaste hubs: Learning from the Israeli-Palestinian e-waste sector. The
Geographical Journal, 185(1), 82-95.

Davis, J. M. & Garb, Y. (2018). A strong spatial association between e-waste burn sites and childhood lymphoma in the West Bank, Palestine. International Journal of Cancer, 144(3), 470-475.

Davis, J. M., Akese, G. & Garb, Y. (2019). Beyond the Pollution Haven Hypothesis: Where and why do e-waste hubs emerge and what does this mean for policies and interventions? Geoforum, 98, 36-45.

Davis, J. M. & Garb, Y. (2017). Participatory shaping of community futures in ewaste processing hubs: Complexity, conflict, and stewarded convergence in a Palestinian context. Development Policy Review, 37(1), 67-89.

Friedlander, Weisbrod N, and Garb (2019). �Climatic and Soil-Mineralogical Controls on the Mobility of Trace Metal Contamination Released by Informal Electronic Waste (e-Waste) Processing.� Chemosphere 232:130�39.

Boxerman, A. �Burning Israeli scrap for valuable raw metals is a lethal livelihood for thousands of Palestinians, sending cancer rates skyrocketing in villages near Hebron.� Times of Israel, 10 January, 2022.

Garb, Y. & Davis, J.M. �Hub-Driven Policy Packages as a Basis for e-Waste Reform: Rationales and a Case Study.� Environmental Politics (January 13,2023): 1�21.

Garb, Y., & Leblond, N. (2024). Flowing toxics: E-waste field work in the Palestinian-Israeli space. Environment and Planning C: Politics and Space, 42(1), 45-63.

Davis J-M, Garb Y. Polluted Politics: The Development of an Israeli-Palestinian E-Waste Economy. Cambridge University Press; 2024. DOI: https://www.cambridge.org/core/books/polluted-politics/C5DBC45480C1DF14252763719BC28799


***************************************************************************
This Readme file template was adapted from that published by the University of Minnesota: 
https://www.lib.umn.edu/datamanagement/metadata
***************************************************************************


